Londei T. 2000. The cheetah (Acinonyx jubatus) dewclaw: specialization overlooked. J Zool , Lond 251:535-47.

Keywords: Acinonyx jubatus/cheetah/claw/dewclaw/Felidae/locomotion/Miracinonyx/morphology/ natural history/predator/prey/puma/Puma concolor/research

Abstract: Both the behavioural and the morphological analyses confirm the special role of the dewclaw in the cheetah: a strong hook to stop running animals by using the energy of the victim itself. However, this would hardly be an explanation for the rather large dewclaw in the puma. The present finding of the puma as an intermediate between the cheetah and other large felids for dewclaw size, supports the idea that, despite some later reversal to a more primitive, typically feline structure, the puma originated from felids like the fossil cheetah-like cat Miracinonyx inexpectatus of North America, which had longer, more cursorial limbs (though maintaining fully retractile claws) and, as far as one can judge from phalanx size, larger dewclaws than the extant puma.

J. Zool., Lond. (2000) 251, 535±547 # 2000 The Zoological Society of London Printed in the United Kingdom

Short communications Edited by M. L. Gorman

J. Zool., Lond. (2000) 251, 535±537 # 2000 The Zoological Society of London The cheetah (Acinonyx jubatus) dewclaw: specialization overlooked

Tiziano Londei Dipartimento di Biologia, UniversitaÁ degli Studi, Via Celoria 26, 20133 Milan, Italy

INTRODUCTION preying on the wapiti Cervus canadensis; 9 instances of cheetah preying on the blue wildebeest, the topi Dama- The cheetah Acinonyx jubatus is an atypical felid, well liscus lunatus, gazelle Gazella granti and G. thompsonii, known for having blunt, only slightly curved, and only and hare. partly retractile claws, clearly an adaptation for high- (2) Measurement of the dewclaw and, for comparison, speed locomotion in the pursuit of swift mammals. the claw of the second digit of the forepaw in museum However, saying that prey `is usually knocked down by specimens, pelts or skeletons. Dorso-ventral height at the force of the cheetah's charge' (Nowak, 1999) is the ungual base was considered as a good indicator of incorrect, because this predator actually relies on the the size of the claws with special reference to their claw of the ®rst digit of the forepaw, the so-called hooking power, and a convenient measure because of dewclaw, to hook the ¯eeing prey off balance. Although frequently worn, broken or, for skeletons, missing (1) this was understood by Indian huntsmen through horny claws. The measure was taken as height of their use of tame cheetahs and published (Burton, 1950) the horny claws that were accessible up to base for pelts, in a natural history journal circulating well outside whereas height of the ungual crest of the distal phalanx India, (2) wounds attributable to the dewclaws of was measured for skeletons. Although some slight effect cheetahs were later observed (Schaller, 1972) on prey of the different methods could not be excluded for the animals in Africa, the area of most research on this absolute values obtained, within-specimen (pelt or ske- felid, and (3) the information from India was ®nally leton) relative values were used to compare the species. reported (though with the wrong year in the citation) in One-way ANOVA with Tukey multiple comparison of an authoritative book on African mammals (Kingdon, the means was performed on the measure of the 1977), no systematic study of the dewclaw itself, or of dewclaw with the measure of the second-digit claw explicitly related questions, has been made. Perhaps the covariant. This procedure met the requirement of a general connotation of the term `dewclaw', as intended parametric variable for ANOVA better than using the for a non-functional claw on a rudimentary digit as in claw ratio, though claw ratios are presented additionally the case of dogs, has masked the importance of what is for easier reference (Table 1). not only a strongly curved and sharply pointed, but also a very large claw in the cheetah. In fact, this specializa- tion of the cheetah has escaped the attention of RESULTS behaviourists (Eaton, 1970), anatomists (Gonyea & Ashworth, 1975), and palaeontologists (Adams, 1979). The predatory sequences showed that tigers, lions and leopards all relied on a strong impact to strike the prey to the ground, though depending on the prey's size the METHODS action varied from a blow with one forepaw to the collision of the predator's entire body. All the claws of The present investigation developed along 2 lines. the forepaws and, in the case of large prey, even of the (1) Frame-by-frame analysis of ®lmed sequences of the hindpaws too, were used by the predator as hooking predatory behaviour of large felids, videotaped from tools to progress with the mouth towards the prey's television. This involved 3 instances of tiger Panthera neck, this being reached very quickly, often before the tigris preying on sambar Cervus unicolor, spotted deer prey fell down. In the one sequence with a puma the Axis axis and Hanuman langur Semnopithecus entellus; predator's jaws seemed less important initially (for prey 4 instances of lion Panthera leo preying on Burchell's grasping), but the body impact seemed even stronger, as zebra Equus burchelli and the blue wildebeest Conno- the puma leaped to the forequarters of the still deer chaetes taurinus; 2 instances of leopard Panthera pardus after an accelerative dash and struck it with both preying on Thompson's gazelle Gazella thompsonii and forearms while taking full grasp with the forepaws. On hare Lepus sp.; 1 instance of puma Felis concolor the contrary, the relative speed of the cheetahs to their 536 Short Communications

Table 1. Comparison of large felids for dewclaw sizea

Species Dewclaw Second-digit claw Claw ratio Dewclaw correctedb Tukeyc

Tiger 28 ‹ 2.0 26 ‹ 2.0 1.1 ‹ 0.006 21 ‹ 0.47 * Lion 30 ‹ 0.6 28 ‹ 0.5 1.1 ‹ 0.008 21 ‹ 0.61 * Leopard 17 ‹ 1.3 16 ‹ 1.1 1.1 ‹ 0.010 21 ‹ 0.35 * Puma 19 ‹ 0.5 15 ‹ 0.3 1.3 ‹ 0.014 24 ‹ 0.42 *

Cheetah 20 ‹ 0.7 14 ‹ 0.4 1.4 ‹ 0.025 26 ‹ 0.45 * a Measures of dorso-ventral height at ungual base, in mm. Mean ‹ se from four specimens of each species. b By second-digit claw covariant. c Multiple comparison of means on dewclaw corrected. Asterisks in same column for homogeneous means at 95% con®dence level.

¯eeing victims was always low, which made the impact Grady & KurteÂn, 1990) that, despite some later reversal rather weak. All the nine sequences showed that only to a more primitive, typically feline structure, the puma when the prey was on its back had the cheetah's jaws a originated from felids like the fossil cheetah-like cat main role, the well-known strangling action. Although Miracinonyx inexpectatus of North America, which had four inexperienced cheetahs bit a standing wildebeest in longer, more cursorial limbs (though maintaining fully various parts of the body during conjunct attacks, the retractile claws) and, as far as one can judge from prey escaped eventually. Also with small prey (hare) a phalanx size, larger dewclaws than the extant puma. cheetah seemed more reluctant than a leopard to use the Therefore, although a more behavioural study may mouth, initially. Irrespective of the prey's size the suggest some special adaptation of the puma's cheetah forced it down through a hampering action, dewclaw, its rather large size may be a leftover trying to anchor one forelimb or both to the prey's character, instead. body. Cheetahs `hanging' on large animals (wildebeest) Although the dewclaw seems to have an important clearly showed dewclaws being their only hooking tools. role in the predatory function of the forepaw in all The analysis of dewclaw relative size (Table 1) met, in felids, and thus it is properly not a `dewclaw', in the part, the behavioural observation. In the tiger, lion and cheetah it seems to have taken on the predatory leopard, the dewclaw is only slightly larger than the function of the entire forepaw, as the other claws lost claw of the second digit, which matches the joint action this function because of locomotory adaptation. A of these (and the other) forepaw claws in predation. The special study of fossil felids with robust ®rst digits, as cheetah clearly shows an enlarged dewclaw, as expected well as of extant felids with large dewclaws, would help from the separate role of this claw. The intermediate to understand the progress towards cheetah dewclaw size of the dewclaw of the puma is an unexpected result, specialization, which has perhaps occurred with less however. extreme outcomes along other lines of felid evolution.

DISCUSSION Acknowledgements

Both the behavioural and the morphological analyses The following people at natural history museums helped con®rm the special role of the dewclaw in the cheetah: a to ®nd suitable specimens for claw measurement: strong hook to stop running animals by using the Giorgio Bardelli and Michela PodestaÁ (Milan), Giuliano energy of the victim itself. However, this would hardly Doria and Primo Gardella (Genoa), Fausto Barbagli be an explanation for the rather large dewclaw in the and Carlo Violani (Pavia), and Paolo Agnelli and puma. The one behavioural sequence analysed in this Stefania Lotti (Florence). study is in line with the habit of the puma making short-distance, surprise attacks (Nowak, 1999). For the massive action of limbs to strike the prey this felid seems REFERENCES closer to representatives of the genus Panthera, these showing no special size or use of their dewclaws. Adams, D. B. (1979). The cheetah: native American. Science 205: When the novel idea of a close relationship between 1155±1158. the cheetah and puma was proposed, their dewclaws Burton, R. W. (1950). The `dew-claws' of the hunting leopard or were not taken into consideration: Adam's (1979) state- cheetah [Acinonyx jubatus (Schreber)]. J. Bombay Nat. Hist. ment that the Old World cheetah lineage had the limbs Soc. 49: 541±543. specialized as `strictly propulsive structures', is inaccu- Eaton, R. L. (1970). The predatory sequence, with emphasis on killing behavior and its ontogeny, in the cheetah (Acinonyx rate because of the predatory dewclaw of the extant jubatus Schreber). Z. Tierpsychol. 27: 492±504. cheetah. The present ®nding of the puma as an inter- Gonyea, W. & Ashworth, R. (1975). The form and function of mediate between the cheetah and other large felids retractile claws in the Felidae and other representative carni- for dewclaw size, supports the idea (van Valkenburgh, vorans. J. Morphol. 145: 229±238. Short Communications 537

Kingdon, J. (1977). East African mammals 3A. London: Academic Chicago Press. Press. van Valkenburgh, B., Grady, F. & KurteÂn, B. (1990). The Plio- Nowak, R. M. (1999). Walker's mammals of the world 1. Baltimore: Pleistocene cheetah-like cat Miracinonyx inexpectatus of North The Johns Hopkins University Press. America. J. Vertebr. Paleontol. 10: 434±454. Schaller, G. B. (1972). The Serengeti lion. Chicago: University of

J. Zool., Lond. (2000) 251, 537±540 # 2000 The Zoological Society of London Printed in the United Kingdom Between-litter siblicide in captive Indian false vampire bats (Megaderma lyra)

Dieter Leippert1, Wolfgang Goymann2* and Heribert Hofer2,3 1 Zoologisches Institut der UniversitaÈtMuÈnchen, Luisenstrasse 14, D-80333 MuÈnchen, Germany 2 Max-Planck-Institut fuÈr Verhaltensphysiologie, D-82319 Seewiesen, Germany 3 Institut fuÈr Zoo- und Wildtierforschung, Alfred-Kowalke-Strasse 17, D-10315 Berlin, Germany

Key words: facultative siblicide, sibling rivalry, fatal sibling aggression, bats, parent±offspring con¯ict

INTRODUCTION mothers may simultaneously nurse offspring from dif- ferent cohorts (Trillmich, 1986, 1990; pers. comm. in Sibling rivalry seems to contradict Hamilton's rule of Mock & Parker, 1997). maximizing inclusive ®tness (Hamilton, 1964). However, when critical resources are scarce and the competitors are close kin the individual must weigh the Fatal sibling aggression in captive false vampire bats direct versus the indirect components of its inclusive ®tness (Mock & Parker, 1997). In facultatively siblicidal During routine observations of a captive colony of species, con¯icts arise mainly because of the phenom- Indian false vampire bats Megaderma lyra, housed in a enon of `parental optimism', in which the parents room of c.15m2 at the University of Munich (see attempt to raise a larger brood than the average amount Leippert, 1994), we observed a probable case of fatal of expected resources allow (Lack, 1954; Temme & sibling aggression. To our knowledge, this is the ®rst Charnov, 1987; Mock & Forbes, 1995). If critical observation of fatal sibling rivalry in a bat, and only resources are low, weaker or subordinate offspring may the second observation of fatal between-litter sibling be killed by a stronger or dominant sibling. Sibling aggression. mortality may result from direct physical damage and/ Indian false vampire bats give birth to a single pup or socially enforced starvation (Mock, 1984). once a year between January and April (Balasingh, Facultative siblicide (aggression between siblings is Subbaraj & Suthakar Isaac, 1994; Goymann, Leippert sometimes so severe that one or more of them are killed; & Hofer, 1999), weaning in captivity occurs after c.3 Mock & Parker, 1997) has been described in many months and sexual maturity is reached after 15 months avian species (reviewed in O'Connor, 1978; Mock, 1984, in males and 19 months in females (Tuttle & Stevenson, 1987; Mock & Parker, 1997) but rarely in mammals. 1982). Fatal sibling aggression in mammals has been docu- False vampire bats were kept for breeding purposes mented in domestic pigs (Fraser, 1990), arctic foxes and fed ad libitum with mice, a natural prey in their Alopex lagopus (Macpherson, 1969), red foxes Vulpes original habitat. The group consisted of three adult vulpes (Henry, 1985), spotted hyenas Crocuta crocuta females (M-010, M-017, and M-009) and one adult male (Frank, Glickman & Licht, 1991; Hofer & East, 1997) (M-016), all caught in caves around Madurai, Southern and Galapagos fur seals Arctocephalus galapagoensis India, in 1991. Except for weekends, bats were routinely (Trillmich, 1986, 1990). Siblicide in Galapagos fur seals observed during their active period through a glass is currently the only known example where the siblings screen for 2±3 h/day. were from different cohorts. Milk production in Additional evidence came from a ®eld study on a Galapagos fur seals is probably limited by maternal colony of free-ranging false vampire bats in Tirunelveli, condition, and maternal condition may vary dramati- India, where individually marked false vampire bats cally from year to year as a result of unpredictable were observed during the breeding season in 1995 (for changes in food resources (El NinÄo). As a result pups details see Goymann et al., 1999). may be weaned between 1 and 3 years of age, and On, 26 January 1993, M-010 gave birth to M-029, a female pup, and nursed it until the middle of May 1993. False vampire teats change colour from red to white *All correspondence to: Wolfgang Goymann, Max-Planck-Institut fuÈr Verhaltensphysiologie, P.O. Box 15 64, D-82305 Starnberg, Germany. when lactation stops (D. Leippert, pers. obs.). M-010's E-mail: [email protected] teats began to change colour on 10 May 1993, and were